The more you impact your readers, the more they will look forword to reading your future articles.

How well you tell a story is as important as the story itself.

BASIC PRINCIPLES

• Technical accuracy: everything is correct.
• Conciseness and clarity: sentences are short and arguments are clearly presented.
• Good organization and structure: in logical order.
• Good English usage: no spelling or grammar errors.
• Usefulness: readers see how to apply the results to their own work.
• Ethicalness: cite key previous work and comparisons are fair.
• Targeted for the audience.

Basic article include: abstract, introduction, methods, results, conclusions, references and appendices.

The introduction and conclusions should be the sections titled in that manner, whereas methods and results should be presented in a few sections whose titles are appropriate for the content of the article.It is useful for the reader if methods and results are not intermixed.Many readers will jump to the results section first, and the better the sections are structured and present a complete picture of the main contributions, the more readers will have an incentive to go back and read the methods section.

BEST STEPS

First, write a few one-line sentences summarizing the points that are most relevant. Then, write a draft outline of the article, with tentative titles for the methods and results sections. For each section, as well as for the introduction, the outline is a set of one-line sentences that summarize the main points to be presented or discussed. Then, write the introduction and the sections by expanding each line of the outline into one or two paragraphs. Read the entire resulting rough draft and rewrite the introduction and edit the other sections.Finally, write the conclusion and then the abstract, and then revise many times to produce a first draft of the entire article. After a few reading and editing passes, we have a draft that can be sent to others for review.

Abstract: a well-written abstract is also important; unlike the conclusion, the abstract has to first quickly position the problem, then say a few words about the methods and finally mention the key results. A common mistake is including in the abstract formulas, citations, and acronyms. Abstracts are published separately, and thus acronyms and citations become meaningless. Plus, such separately published abstracts are usually in databases that support only regular text, and hence formulas cannot be correctly rendered.

Conclusions: good conclusions section is important. It is your opportunity to clarify for the reader the main contributions of the paper and the most important results. It’s where you present the punch lines. Otherwise, readers will make their own conclusions, which will most likely not match your own.

THE AUDIENCE

The more you think as the reader, the easier it will be to figure out ways to convey your points as clearly as possible.

Your readers usually cluster into four main groups: 1) reviewers, who have a keen eye to find fallacies in your arguments; 2) subject experts, who want to skip the introduction and quickly jump into the main contributions; 3) non-experts, who work on related subjects and selected your paper as a step for broadening their knowledge; and 4) those who don’t know much about the subject, but stumbled on your paper and got curious enough to read it. Except for tutorial papers, usually your main target is the expert reader (groups 1 and 2). Even then you should not miss opportunities to reach other readers (groups 3 and 4) for a broader impact.

In broadening the reach of the article, the introduction, abstract, and conclusions play a key role. Thus, to write them effectively, one good tip is to take a good break after finishing the methods and the results sections before you start working on expanding the introduction outline into paragraphs.

THE TIMING

It is important to pace your article appropriately. The introduction has to sell the importance of the problem, your idea, and your results to entice the reader to keep reading. It will most efficiently do so if it is as clear and short as possible.

Also important is spending enough space on the results section. Many articles make the mistake of spending too much space on the methods sections, thus discussing the results sometimes with just a statement like “It is clear from Table 1 that our proposed method is a significant improvement over others,” to which the reader reaction is likely to be “Huh?” What does “significant” mean? Is the improvement worth the effort of considering this new idea? Does it come with penalties in other relevant metrics? Sometimes a small improvement can be a big deal in practice, as long as it’s correctly positioned. I’ll happily take a 5% reduction in operation costs for a data center, for example, but I won’t bother with a new image encoding algorithm that improves compression ratio by 10% at the expense of a 2 increase in computation.

COMMON ERROR

• DON’T overstate your points and conclusions.
• DON’T forget to mention competing results and other clearly related approaches found in the literature.
• DON’T “forget” to discuss the shortcomings of your ideas; readers will eventually find them out, and it’s better to hear them from you.
• DON’T rush in writing the introduction, abstract, and conclusions.
• DON’T say that your great idea A is much better than your naive idea B; position A with respect to existing literature.
• DON’T be confused about the meaning of words (it’s versus its, affect versus effect, since versus because, which versus that, principal versus principle, etc.)
• DON’T use illustrations with text that are too small to read; reducing figures to illegible size is NOT a good way to gain space.
• DON’T overexplain background ideas, concepts, or equations that can be replaced by good citations.
• DON’T assume your reader is an expert and knows all acronyms.
• DON’T end an abstract by saying “performance will be discussed;” use instead a punch line: “Our method improves A by X %.”
• AVOID qualifiers such as very, intense, little, etc; use quantitative arguments instead.
• AVOID pompous writing (say use, not utilize; end, not terminate; find out, not ascertain; need, not necessity); simple is better.
• AVOID cliché terms such as synergy, paradigm, proactive, top notch, world class, etc.
• DO plan ahead; write and revise a good outline before you start writing paragraphs.
• DO write the methods and results sections before expanding the introduction outline and writing the abstract and conclusions.
• DO use figures and illustrations, but as few as possible, so each can be large enough to be read without difficulty.
• DO focus on results, not on methods; e.g., instead of saying “We use machine learning techniques to study ways to improve A,” say “We get an X % reduction in error using machine-learning technique Z.”
• DO follow the composition rules in [1]–[6].
• DO spend extra time at the end to refine the introduction, abstract, and conclusions.
• DO use the active voice; it strengthens actions and thus the argument. For example, instead of saying “the input signal is processed by the preconditioning filter A,” say “filter A removes high-frequency noise from the input signal;” this eliminates the redundant “is processed” and adds the reason for filtering.
• DO revise and edit; cut, cut, cut!
• DO use spelling and grammar checking.
• DO spend time to make your illustrations visually pleasing.
• DO use a direct, factual tone; breezy or opinionated descriptions are distracting.
• DO break long sections into subsections; subsection titles should be brief and draw attention to key points.
• DO use the conclusions section to emphasize the main contributions of your research; don’t include a summary of what was described in the methods section.
• DO use certain terms with care; say “to optimize performance” or “to maximize usefulness” only if the system does indeed maximize an appropriate metric.
• At many stages during the writing process, take a break, “step out of the paper,” and think like a reader. Reread what you wrote with a critical eye to find parts that are unclear, unjustified, or not stressed well enough.

BEST PRACTICES

Keep reread the article with questions in mind such as:“Did I overstate my arguments?,” “Can I make this point in a more concise and direct manner?,” “Is this figure really necessary?,” “Is this word really needed?,”.